Investigation into geotechnical properties of clayey soils...

Scientia Iranica A (2019) 26(3), 1122{1134

Sharif University of TechnologyScientia Iranica

Transactions A: Civil Engineeringhttp://scientiairanica.sharif.edu

Investigation into geotechnical properties of clayey soilscontaminated with gasoil using Response SurfaceMethodology (RSM)

F. Mir Mohammad Hosseini, T. Ebadi�, A. Eslami, S.M. Mir Mohammad Hosseini,and H.R. Jahangard

Faculty of Civil & Environmental Engineering, Amirkabir University of Technology, Tehran, P.O. Box 15875-4413, Iran.

Received 31 December 2016; received in revised form 25 July 2017; accepted 21 October 2017

KEYWORDSClayey;Sandy clay;Gasoil;Laboratory tests;Mechanical properties;Response SurfaceMethodology (RSM)

Abstract. Oil and its derivatives such as gasoline, motor oil, and gasoil are being used invarious industrial and non-industrial sectors as the main energy sources all over the world.Unfortunately, in the processes of exploration, transportation, and storage, they may spillor leak into the soil. Among them, gasoil, which is more widely used in di�erent partsand machineries, has the largest contribution to contamination of the lands. Purgationof these areas is not always feasible or possible. Instead, they can be used in manyengineering practices if the level of contamination is not high. In such cases, knowingthe geotechnical properties of these areas is of great necessity and importance. In thisstudy, extensive laboratory tests were performed on remolded clayey samples mixed withgasoil to evaluate their geotechnical properties. The Response Surface Methodology (RSM)was used to analyze the data and �nd behavioral equations. According to testing resultsand RSM outputs, decrease in Atterberg limits and increase in maximum dry density occurby increasing contamination. Also, both of the shear strength parameters (c and f) exhibita turning point at 8% gasoil content, while their variation trends are quite in oppositedirections.© 2019 Sharif University of Technology. All rights reserved.

1. Introduction

Besides the air and water pollution, soil contaminationhas been studied due to its signi�cant e�ects on theenvironment. Soil contamination can be studied fromtwo points of view, one is the release rate of infectionand its spread in the ground, and the other is its

*. Corresponding author. Tel.: 0098-21-64543031E-mail addresses: fateme [email protected] (F. MirMohammad Hosseini); [email protected] (T. Ebadi);[email protected] (A. Eslami); [email protected] (S.M.Mir Mohammad Hosseini); [email protected](H.R. Jahangard)

doi: 10.24200/sci.2017.4574

e�ects on soil properties. Di�erent site pollutions withmaterials and petroleum products occur for variousreasons. The leakage and spillage of gasoil from old andcorroded storage tanks, pipelines, processing plants,and petroleum transportation facilities contaminatethe surrounding soils. The extent of contaminationdepends on the �ltration and retention properties ofthe soil [1-3].

Contamination due to chemicals and oil spills caninuence the engineering behavior of soils. Severeenvironmental and ecological problems are caused byoil spills. They may also have adverse e�ect onengineering properties of soil such as shear strength,compressibility, and hydraulic conductivity. The e�ectof oil contamination on the engineering properties

F. Mir Mohammad Hosseini et al./Scientia Iranica, Transactions A: Civil Engineering 26 (2019) 1122{1134 1123

of soil has been increasingly focused on in recentyears. Investigations have shown that the engineeringproperties and behavior of soil may be signi�cantlyinuenced by chemical contamination [4-9].

It is recognized that remediation of soil is in prac-tice costly and limited, especially in widely contami-nated areas, due to the huge expenses demanded. Analternative solution could be the use of contaminatedsoil in engineering practices, such as embankments,road bases, back�lls, etc. Thus, in addition to theenvironmental concerns about groundwater pollutionand other possible e�ects, an investigation into thegeotechnical characteristics of the contaminated soilis required. The investigation has also to be done todesign a storage tank foundation on contaminated soilsin a way to have satisfactory performance during thelifetime of the tank. The investigation also providesknowledge to revise the foundation design of the exist-ing structures on contaminated sites.

A number of studies have already been carriedout on the geotechnical properties of the soils contam-inated by petroleum hydrocarbons [10-13]. Al-sanadet al. [14] and Al-sanad and Ismael [15] investigatedthe geotechnical properties of Kuwaiti oil contaminatedsand. They carried out an extensive laboratory testingprogram to study the basic properties by doing com-paction, permeability, triaxial, and consolidation tests.Contaminated specimens were prepared by mixing thesand with oil with the amount of 6% by weight. Theresults indicated a small reduction in strength andpermeability and an increase in compressibility due tocontamination.

Khamehchiyan et al. [16] performed a labora-tory testing program to determine e�ects of crude oilcontamination on geotechnical properties (includingAtterberg limits, permeability, uniaxial compressivestrength, direct shear tests, and compaction charac-teristics) on three soil types of CL, SP, and SM inthe south of Iran. The results indicated decrease instrength, permeability, maximum dry density, opti-mum water content, and Atterberg limits.

Puri [4] studied the geotechnical aspects of con-taminated sands and e�ects of oil contaminationon compaction characteristics, shear strength, one-dimensional compression, and hydraulic conductivity.Adverse e�ects of oil contamination on shear strengthparameters and 20-25% reduction in the friction angleof sand contaminated with oil were observed in thestudy.

Nazir [17] conducted a laboratory testing pro-gram to study the e�ect of motor oil contaminationtogether with the e�ect of contamination duration ongeotechnical properties of over-consolidated clay. Fourparameters of Atterberg limits, uncon�ned compressivestrength, coe�cient of permeability, and compress-ibility characteristics were investigated. The studies

revealed that the uncon�ned compressive strength wasreduced by about 38% as compared to the referencevalue (uncontaminated).

A comprehensive set of laboratory tests wereconducted on both uncontaminated and contaminated�ne-grained soils containing di�erent amounts of crudeoil by Kermani and Ebadi [18]. The results indicatedan increase in the angle of internal friction, maximumdry density, compression index, and Atterberg limitsas well as a decrease in optimum water content andcohesion, as the oil content increased. Khosraviet al. [11] evaluated the geotechnical properties ofgasoil-contaminated kaolinite. They carried out anextensive laboratory program that studied basic prop-erties, namely Atterberg limits, consolidation, directshear, and uncon�ned compression tests in clean andcontaminated kaolinite specimens. Results indicatedan increase in the cohesion and decrease in both thefriction angle and compressibility of kaolinite soils withincreasing the gasoil content.

More recently, Liu et al. [19] focused on engineer-ing properties of kaolin clay contaminated by diesel oil.They carried out comprehensive tests on clay sampleshaving oil contents of 4, 8, 12, 16, and 20% (by massfraction). They found that as contamination degree ofthe kaolin clay increased, both the liquid and plasticlimits decreased, but there was only a slight increasein plasticity index. They stated that the Uncon�nedCompressive Strength (UCS) of the oil contaminatedkaolin clay was inuenced by not only oil content butalso curing period. They concluded that the oil contentof 8% was the critical value for engineering propertyof the kaolin clay to transit from water-dominatedtowards oil-dominated characteristics.

In the early 2016, Nasehi et al. [20] studiedthe inuence of gasoil contamination on geotechnicalproperties of only granular (�ne and coarse) soils. Theyconducted some laboratory tests, such as plasticity,compaction, Uncon�ned Compressive Strength (UCS),and direct shear, on uncontaminated and arti�ciallycontaminated specimens with 3, 6, and 9% gasoilrelative to dry weight of some SP and ML samples.The results indicated that a decrease in the frictionangle and an increase in the cohesion of the soil withincrease in gasoil content would occur. In addition,reduction in the maximum dry density and optimummoisture content was observed during the conductionof compaction tests. The increase in gasoil content alsoshowed a direct e�ect on the increase in liquid andplastic limits of the silt samples.

Regarding the investigations performed so far, re-searchers have less considered the gasoil as contaminantamong the petroleum products. However, gasoil as oneof the most common oil derivatives in many enginesand systems, such as transport, industry, agriculturalmachinery, and diesel engines, has an important con-

1124 F. Mir Mohammad Hosseini et al./Scientia Iranica, Transactions A: Civil Engineering 26 (2019) 1122{1134

Table 1. XRD analysis of the kaolinite.

Chemical components L.O.I SiO2 Al2O3 Fe2O3 TiO2 Na2O K2O CaO MgO

Percentage (%) 8.9 64.4 21.2 0.26 0.05 1.11 1.15 2.2 0.73

tribution to the pollution of the soils. Clayey soils, dueto their fabrics and nature, may face more complicatedand inuential changes in their engineering propertiescompared with non-cohesive soils such as what wasrecently studied by Nasehi et al. [20]. However, in thepast investigations, limited studies have been carriedout even on granular and sandy soils.

In order to model the behavior of contaminatedclayey soils, analyze the geotechnical data from lab-oratory tests, and �nd correlations of parameters,an appropriate method should be used. Factorialexperimental design in combination with ResponseSurface Methodology (RSM), as a statistical analysisapproach, is an e�cient and widely used method toanalyze, compare, and optimize the simultaneous ap-plication of di�erent factors [21]. The Response SurfaceMethodology (RSM) is a combination of mathematicaland statistical approaches used for modeling di�erentphenomena and optimizing the experimental results asa function of various parameters [22]. The method,which has been used by many researchers, requires alimited number of experiments, being a simple andfeasible optimization procedure.

In the current study, the e�ort has been at-tempted to investigate the geotechnical properties of�ne soils contaminated by gasoil. A series of physicaland mechanical tests were carried out on clayey soilsamples mixed with di�erent percentages of gasoil. Toanalyze the experimental data and extend the resultsfor getting a general behavioral model indicating theengineering properties of �ne soils contaminated bygasoil, the Response Surface Methodology (RSM) wasused. The innovations of the current study can beclassi�ed as follows:

Investigating the e�ect of gasoil contamination onthe following geotechnical parameters of clayey soils:

{ Atterberg limits (P.L., L.L., and P.I.),

{ Compactability of the soil (wopt. and gdmax),

{ Shear strength parameters (C and f),

{ The role of non-cohesive �ne contents.

Developing experimental equations to predictchanges of the above parameters using RSM.

The details of testing program, analyzing approach,and the main outcomes of this study are described inthe following sections.

2. Testing materials and methods

2.1. MaterialsSoil materials of this study included pure clay andsandy clays having di�erent �ne sand contents. Theclay which was the main part of the studied soil waskaolinite and in the process of performing the study,some percentages of �ne sand were added to it inorder to determine and compare the e�ects of gasoilcontamination on engineering properties of both pureand sandy clays. X-Ray Di�raction (XRD) was done toidentify and analyze the composition of the clay. Theresults are shown in Table 1. Table 2 summarizes theclay mineralogy. The grain size distribution of the clay,determined by a hydrometer (ASTM-D422), is shownin Figure 1. The sand is Firozkooh arti�cial �ne sand(speci�ed by grade No. F161). Tables 3 and 4 presentthe characteristics of the soils used in this research andFigure 2 shows the grain size distribution of the �nesand (ASTM-D422). The pure clay and �ne sand usedin this study are separately classi�ed as low plasticclay (CL) and poorly graded sand (SP), respectively,based on the uni�ed soil classi�cation system (ASTM-D2487). Table 5 shows the gas oil properties used inthe current study.

2.2. Response Surface Methodology (RSM)Factorial design and response surface methodologywere used for experimental design. Based on the

Table 2. Clay mineralogy.

Kaolinite Feldespar Silicon51.57% 16.22% 32.21%

Table 3. The clay properties and indices.

PL(%)

LL(%)

PI(%)

Gsgdmax

(g/cm3)wopt(%)

16 48 32 2.66 1.46 28.84

Figure 1. Grain size distribution curve of the kaolinite.


Table 4. Firozkooh �ne sand (F161) speci�cations.

D10 (mm) D30 (mm) D50 (mm) D60 (mm) Gs emin emax Cu Cc0.16 0.20 0.25 0.26 2.65 0.58 0.931 1.625 0.96

Table 5. The gasoil properties (NIORDC, Iran).

Density in 150�C F.B.P (max) Flash point (min)(kg/L) (�C) (�C)

0.820-0.860 385 54

Figure 2. Grain size distribution curve of the �ne sand.

information obtained from the pre-trial and studies forstatic tests, 2 parameters were examined as operatingparameters and their boundaries determined. Thesetwo parameters were the sand (A or x1) with the rangeof (0-20%) and gasoil (B or x2) with the range of (0-16%) as independent variables. They were coded atthree levels between 1 and �1. The ranges of theindividual factors were chosen based on the preliminarystudies and previous experience, and are presented inTable 6. The Design Expert 7 software was used forstatistical analysis of experimental data by responsesurface methodology. Based on the number of theselected parameters, the software determined 13 runsfor the tests consisting of 8 non-central points and 5replicates at the center according to Central CompositeDesign (CCD). For the statistical calculations, the vari-ablesXi (the actual value of independent variable) werecoded as xi (dimensionless value of the independentvariable) according to the following equation [22,23]:

xi = (Xi �X0)=�X; (1)where X0 is the Xi value in the central point and �Xrepresents the step change. The experimental datawere �tted into the empirical second-order polynomialmodel. The second-order equation to predict theoptimal condition is [22,23]:

Y = b0 +nXi=1

bixi +nXi=1

biix2i +n�1Xi=1

nXj=i+1

bijxixj ;(2)

where Y is the predicted response by the model, b0is a constant, bi are linear coe�cients, bii are second-order coe�cients, bij are interaction coe�cients, and xi

Table 6. Experimental design and the levels ofindependent process variables.

Independentvariables

Symbol Coded levels-1 0 1

Sand A 0 10 20Gasoil B 0 8 16

Table 7. Composition of prepared and tested samples.

Number Sample1 Pure clay + 0% gasoil2 Pure clay + 4% gasoil3 Pure clay + 8% gasoil4 Pure clay + 12% gasoil5 Pure clay + 16% gasoil6 Clay + 10% sand + 0% gasoil7 Clay + 10% sand + 4% gasoil8 Clay + 10% sand + 8% gasoil9 Clay + 10% sand + 12% gasoil10 Clay + 10% sand + 16% gasoil11 Clay + 20% sand + 0% gasoil12 Clay + 20% sand + 4% gasoil13 Clay + 20% sand + 8% gasoil14 Clay + 20% sand + 12% gasoil15 Clay + 20% sand + 16% gasoil

and xj are the coded values corresponding to the testedvariables [23,24]. The validity of the predicted model isveri�ed by analysis of variance (ANOVA). The second-order model quality is assessed by the correlation factor(R2) and analysis of the result is carried out usingFischer trial and probability value with 95% con�dencelevel [22].

Predictive experiments for static tests, includingthe Atterberg limits (ASTM-D4318) and standardproctor tests (ASTM-D698), were performed to deter-mine the physical indices and maximum dry density.Moreover, triaxial (ASTM-D4767) and direct shear(ASTM-D3080) tests were carried out to get the shearstrength parameters. The details of samples preparedand tested are given in Table 7. For the software inputresponses, 2 key responses from each test were chosen.Maximum dry density and optimum water content werethe 2 responses of the compaction test. Atterberglimits have Plastic Limit (PL) and Liquid Limit (LL)as the responses. Direct shear and static triaxial tests


also have 2 responses, which are cohesion and frictionangle. The experimental tests, which were determinedby the software, were done in soil mechanics laboratoryof Amirkabir University of Technology. Then, therequired responses for the tests were entered into thesoftware in order to analyze and achieve an appropriatemodel.

3. Sample preparation method

As water content of the pure clay soil used in thisstudy was zero percent, the soil did not require to bedried in the furnace and was only pulverized in orderto pass through a size-4 sieve. Then, the gasoil wasadded to the soil samples with the amounts of 0, 8,and 16% the weight of the dry soil (the combinationand percentages of the materials were selected by thesoftware). On the other hand, to make the sandy claysamples, the Firozkooh �ne sand was mixed with pureclay in di�erent percentages up to 20% the weight of thedry kaolinite. Accordingly, 15 samples were preparedfor laboratory tests. For making samples contaminatedwith gasoil, the soil was mixed and stirred with gasoiluntil getting a homogenous soil-gasoil mixture. Thesame procedure was followed to prepare sandy claysamples (with or without gasoil). Then, all thespecimens were housed in fully sealed plastic bags andkept in the temperature of 30�C for 7-10 days for curingwithout any gasoil being evaporated. This period oftime is consistent with the 3-7 days period proposedin the literature for soil-contaminant mixtures [11,25].The temperature was chosen based on the averagetemperature within a reasonable depth in the oil facilityand sites prone to similar contaminations [11]. Allthe specimens were prepared in their maximum drydensities and optimum water contents based on theresults of compaction tests (as shown in Figure 3) wereinitially used.

4. Testing results

After 15 runs, based on a factorial design of twoindependent variables at three levels, the experimentalresults for the tests were obtained and the modelanalysis process began with analysis of variance. TheANOVA table and the resulting equation represent theprediction of the model by the software for geotechnicalproperties of soil parameters. One of the data ofANOVA by which the validity of the model is measuredis the p value. P value is the probability of a givendata with void hypothesis. The lower the amount of pvalue (< 0:05), the more the validity and importanceof the model will be. In this study, in order to validatethe models predicted by the software, the coe�cientR2s, analysis of variance, and normal probability plotsfor the remaining graphs will be investigated. The

Figure 3. Results of compaction tests carried out ondi�erent contaminated samples.

values close to unity for coe�cient of determinationindicate the high validity of the model. In the normalplot of residuals, the closer the points to a straightline, the more normal the distribution will be. Theresults for these models as well as analyses of variancesand equations are provided for the experiments in thefollowing.

4.1. Atterberg limitsThe LL (Liquid Limit), PL (Plastic Limit), and PI(Plastic Index) were measured and evaluated for 15samples of clean and contaminated soils with 0, 8,and 16% gasoil by the weight of dry soil. In case ofgasoil-contaminated soils, the regular equation of thewater content in the soil, which is the ratio of weightof water to the weight of dry soil, due to the presenceof gasoil and its e�ect on the uid amount inside thevoids, cannot be used. In order to investigate the e�ectsof gasoil on the water content, an extensive laboratorystudy was done. Fifty specimens with di�erent contentsof soil, gasoil, and water were prepared. The specimenswere classi�ed in two groups. In both groups, thesoil and water contents were both quite the same,changing in the ranges of 20-300 grams and 20%-50%,respectively. The only di�erence between these groupswas that there was no gasoil content in the �rst one,while in the second one, the specimens were mixed withgasoil in the range of 4-20%. Thus, after taking thesamples out of the oven, each specimen from group 1could be compared with its equivalent from group 2 formeasuring the amount of gasoil loss in the oven for theknown amounts of soil and initial water content. Sincethe F.B.P. (Fuel Boiling Point (see Table 3)) of the


gasoil was considerably greater than that of the water,the amount of gasoil evaporation, after the period ofcomplete evaporation of water in the samples in theoven, was not too great.

It was found that gasoil evaporation was a func-tion of three factors, namely gasoil percentage, amountof soil, and water percentage, in the sample. Themore the gasoil percentage, the higher its evaporationrate will be. The gasoil that does not evaporate willremain in the soil, unlike the water the whole amountof which will evaporate. It also follows that the higherthe amount of soil, the lower the gasoil evaporation inthe soil will be. In case of di�erent water contents, nomonotonous trend was achieved. Therefore, two cor-rection factors of soil weight and water percentage wereapplied in calculation of gasoil evaporation. Then, thereal water contents were calculated for the specimensby considering the gasoil evaporation e�ects (real watercontent = moisture content of the specimen � gasoilevaporation content) by means of some graphs andequations. Figure 4 shows the correlation between sandand gasoil contents with gasoil evaporation. Increasein gasoil content leads to increase in gasoil evaporation.

Sand content reduction has almost the same e�ecton gasoil evaporation. Figure 5 indicates the relationof water contents in clay and gasoil contents with gasoilevaporation.

Based on the above corrections, the Atterberg

Figure 4. Relationship of gasoil evaporation with sandand gasoil contents.

Figure 5. Relationship of gasoil evaporation with waterand gasoil contents.

limits were calculated. As mentioned before, 2 re-sponses (PL and LL) were chosen for Atterberg limitstest. Tables 8 and 9 are ANOVA tables for the �rst andsecond responses of Atterberg limits. For variables,p values lower than 0.05 indicate that the modelterms are signi�cant, while values greater than 0.1000indicate that the model terms are not signi�cant [26].In addition, lack of �t p value may possibly be due toa sentence in the model with void assumption, not thewhole model. One of the most important factors thatcontrols model veri�cation is the amount of lack of �t pvalue, which should be considered unimportant by themodel. This means that the model has been validatedand performs well. The PL and LL responses were�tted to the second order polynomial (Eqs. (3) and (4))in terms of actual factors:

PL = + 32:06� 0:295A� 0:204B + 0:026A:B� 8:12E � 3�A2 � 0:19B2; (3)

LL = + 47:88� 0:474A+ 0:052B + 0:036A:B� 0:010A2 � 0:049B2; (4)

where PL is Plastic Limit (%), LL Liquid Limit (%),A the �ne sand content (%), and B the gasoil content(%).

In Tables 8 and 9, the sentences A and B arevery impressive in the model (� 0:05). The lack of

Table 8. ANOVA for response surface quadratic model for plastic limit.

Source Sum of squares df� Mean square F value p valueProb > F

Model 65.46 5 13.09 689.53 < 0:0001A: sand (%) 18.63 1 18.63 981.03 < 0:0001B: gasoil (%) 12.88 1 12.88 678.17 < 0:0001AB 2.49 1 2.49 131.16 0.0003A2 0.88 1 0.88 46.36 0.0024B2 2.13 1 2.13 111.94 0.0005Residual 0.076 4 0.019Lack of �t 0.024 1 0.024 1.39 0.3231

R2 = 0:9988, Adj�R2 = 0:9974, Pred�R2 = 0:9854.�df = degrees of freedom (the number of model sentences minus one):


Table 9. ANOVA for response surface quadratic model for liquid limit.

Source Sum of squares df� Mean square F value p valueProb > F

Model 157.73 5 31.55 457.11 < 0:0001A: sand (%) 44.78 1 44.78 648.80 < 0:0001B: gasoil (%) 27.42 1 27.42 397.34 < 0:0001AB 5.08 1 5.08 73.67 0.0004A2 1.52 1 1.52 21.96 0.0054B2 14.54 1 14.54 210.64 < 0:0001Residual 0.35 5 0.069Lack of �t 0.077 1 0.077 1.15 0.3439

R2 = 0:9978, Adj�R2 = 0:9956, Pred�R2 = 0:9798.�df = degrees of freedom (the number of model sentences minus one).

Table 10. Full factorial CCD matrix for liquid limit.

Run orderFactors

Actual value Predicted value ResidualA: sand (%) B: gasoil (%)5 0 8 48 47.9 0.14 10 8 38.5 38.2 0.37 10 8 41 40.5 0.511 20 0 34 34.1 -0.11 0 0 45 45.1 -0.110 10 8 37.5 37.4 0.112 0 16 42 42.1 -0.113 10 8 36.18 36.1 0.18 10 8 42.6 42.3 0.32 20 8 42.5 42.3 0.29 20 16 42.1 42.3 -0.23 10 0 42 42.3 -0.36 10 16 42.2 42.3 -0.1

Figure 6. Response surface plot of the Plastic Limit(PL).

�t p values for the two models is insigni�cant from theviewpoint of the software. Table 10 shows an exampleof the predicted and actual values for the liquid limit.Response surface and contour plots of PL and LL arepresented in Figures 6 and 7, respectively.

4.2. Density and compaction characteristicsTo study the e�ects of gasoil on compaction behaviorsof clay and sandy clay soils, standard proctor com-paction tests were carried out on the selected samples.

Figure 7. Contour plot of the Liquid Limit (LL).

For compaction test, 2 responses of maximum drydensity and optimum water content were investigated.It is worth mentioning that to �nd the actual moisturecontent in the presence of gasoil, weight of evaporatedgasoil was subtracted from moisture content of thewhole specimen by knowing the weight of evaporatedwater. ANOVA tables for the �rst and second re-sponses of compaction test were also prepared. The


main factors (A and B) are very important and, basedon the values obtained from the ANOVA tables, bothshow acceptable ranges (� 0:05). Eqs. (5) and (6)have been developed in terms of actual factors. Theseequations show the relationship between maximum drydensity and optimum water content with sand andgasoil contents:

gdmax = + 1:46 + 2:66E � 3A+ 5:35E � 4B+ 2:50E � 4AB + 1:48E � 4B2; (5)

wopt = + 24:91� 0:247A� 0:435B � 2:06E � 3AB+ 6:20E � 3A2 � 0:014B2; (6)

where gdmax is maximum dry density (g/cm3), woptoptimum moisture content (%), A the �ne sand content(%), and B the gasoil content (%).

The lack of �t p values for the two models isinsigni�cant. Figure 8 also indicates the normal plot ofresiduals for maximum dry density. Response surfaceand contour plots of water content and maximum drydensity are presented in Figures 9 and 10, respectively.

Figure 8. Normal plot of residuals for maximum drydensity.

Figure 9. Response surface plot for the optimum watercontents.

Figure 10. Contour plot of maximum dry density of thespecimens.

4.3. Shear strength parameters4.3.1. Triaxial testsIsotropically consolidated undrained triaxial tests wereperformed on samples. All samples were 50 mm in di-ameter and 100 mm in height, and they were preparedat a relative density of 90% with the correspondingwater content. The reason for this selection was thatall samples showed dilative behavior and they were tobe placed into the dry side of the compaction curve andfar from saturated state. After mixing the soil with thedesired contents of water and gasoil, the samples werekept in sealed plastic bags for 24 hours prior to thetest. Then, for reaching the desirable compaction, wettamping method was used. The tamping rod had thediameter of 2.5 cm, which was one half the diameter ofthe specimen. In order to plot Mohre circles and obtainthe shear strength parameters, each test was done forthree con�ning stresses of 100, 200, and 300 KPa.The slope of failure envelop of the circles is internalfriction angle and its intercept is cohesion. For the testdata, ANOVA tables for the angle of internal frictionand cohesion of the samples are also presented. Bothof the main factors (A and B), based on the valuesobtained from the ANOVA tables, show acceptableranges (� 0:05). Eqs. (7) and (8) show the relation ofinternal friction angle and cohesion of the soil sampleswith �ne sand and gasoil contents:

f = + 26:074 + 0:139A� 3:31B � 3:14E � 3AB� 2:34E � 3A2 + 0:190B2; (7)

C = + 97:84� 0:491A+ 11:71B + 8:34E � 3AB+ 0:014A2 � 0:904B2; (8)

where f is the angle of internal friction (degree), C thecohesion (g/cm2), A the �ne sand content (%), and Bthe gasoil content (%).


The lack of �t p values for the two models is in-signi�cant from the viewpoint of the model. Figure 11also indicates the normal plot of residuals for internalfriction angle. Response surface and contour plots ofcohesion and internal friction angle are presented inFigures 12 and 13, respectively.

4.3.2. Direct shear testsDirect shear tests were done on the selected samples.The tests were performed in a square shear box(10 cm � 10 cm). Rate of shear was 1 mm/min at

Figure 11. Normal plot of residuals for friction angles.

Figure 12. Response surface plot of the cohesion.

Figure 13. Contour plot of the friction angle.

normal loads of 10, 20, and 30 kg. Again, the ANOVAtables for the two main factors (cohesion and frictionangle) based on the testing data are presented. Theresponses were �tted with the second order polynomial(Eqs. (9) and (10)) in terms of actual factors:

f = + 28:82 + 0:638A� 2:69B � 1:56E � 3AB� 0:018A2 + 0:159B2; (9)

C = + 0:616� 0:015A+ 0:023B + 4:062E � 4AB2:94E � 4A2 � 2:22E � 3B2; (10)

where f is the angle of internal friction (degree), C thecohesion (g/cm2), A the �ne sand content (%), and Bthe gasoil content (%).

In the ANOVA table for the internal frictionangle, A, B, and B2 and in the ANOVA table for thecohesion, A and B2 are very important sentences in themodels. Response surface and contour plots of internalfriction angle and cohesion are presented in Figures 14and 15, respectively.

Figure 14. Response surface plot of the friction angle.

Figure 15. Contour plot of the cohesion.


5. Discussion

5.1. Atterberg limitsAs it can be seen in the response surface plot of PL andcontour plot of LL (Figures 6 and 7), the results showa decrease in limits with increasing gasoil and sandcontents in specimens. This reduction in Atterberglimits can be explained by double-layer theory ofclays. The interaction of the short-range repulsion forcewith the long-range attraction between clay particlesmay have a profound inuence on the engineeringproperties of the soil [27]. The compressibility ofkaolinite with non-polar uids like gasoil is highlysensitive to the value of the dielectric constant; doublelayers cannot develop in non-polar uids [28]. Theliquid limit values decrease consistently with decreasein the dielectric constant of the pore uid [29,30]. Thetwo forces that act between the clay particles are thedi�use double layer repulsion and the van der Waalsattraction [30]. Decreasing dielectric constant of thepore uid compresses the di�use double layer thicknessaround the clay particles and in turn, the electrostaticrepulsion, which can cause coagulation [30,31]. Thus,when the gasoil is mixed with soil, it surrounds thesoil particles and then, the water reaction with soilparticles is reduced. As a result, the thickness ofdouble-layer water decreases [16]. A decrease in At-terberg limits occurs in the gasoil contaminated soil.It can also be seen that the liquid limit depends onphysicochemical factors and, to a lesser degree, onmechanical factors other than the pore uid density.Meegoda and Ratnaweera [32] studied the mechanicaland physicochemical factors that controlled liquid limitof soil. They stated that if water was used as pore uid,the inuence of mechanical factors would remain thesame. However, if an organic uid was used instead ofwater, then the physical properties of the uid suchas viscosity would inuence the liquid limit. Thephysicochemical factors due to low dielectric constantvalues would cause the clay to behave more like agranular material, in the presence of oil contamination,thus lowering the liquid limit [16]. These results areconsistent with the studies of Khamehchiyan et al. [16]and Olgun and Y�ld�z [30]. As it can be seen, the morethe sand content, the less will be the Atterberg limits.It is because sand makes the texture of the samplesmore granular.

5.2. Density and compaction characteristicsIn Figures 9 and 10, as the gasoil content increases, themaximum dry density increases and the response sur-face gradually decreases with optimum water content.In sandy clay specimens, increasing the percentage ofsand to pure clay also increases the maximum drydensity and decreases the optimum water content.This improvement in compaction characteristics can

be attributed to the lubricating e�ect of the gasoil,which is due to the oil �lms coating on the individualclay particles and the clay groups [18,33]. In addition,as the results of Atterberg limits tests represent, thesoil plasticity decreases due to gasoil contamination.A reduction in soil plasticity may lead to better com-paction. Compactability is an important parameter forsoils that are used as paving materials. This changein properties enhances the use of soil in embankmentsfor road or other geotechnical structures [18]. Thegasoil has lubricating e�ect and decreases the frictionbetween soil particles. Thus, gasoil content incrementincreases the compaction e�ciency because of thehigher lubrication. It means that with a certain amountof energy consumption, higher maximum dry densitywill be obtained with lower optimum water content,because a small amount of gasoil has more inuentialrole than water in friction reducing process. Fine sandincrease has the same e�ect as the gasoil increase onthe maximum dry density and optimum water contenton pure clay samples; however, when the two frictionreducing factors (sand particles and gasoil) are addedto the clayey soils, the increase in the maximum drydensity and decrease in the optimum moisture contentwill be doubled. As a result, increase in the intensityof the compression e�ciency is achieved. It seems thatin case of adding sand to the pure clay, the e�ciencyis more than that for the pure clay. However, thisis not the optimal condition yet. In gasoil contentsless than 8%, the e�ects cannot be ampli�ed; but ingasoil contents more than 8%, sand particles can bemore inoculated with gasoline. Thus, the e�ects can beintensi�ed and the e�ciency of the compaction e�ortwould be the most.

5.3. Triaxial testsTypical results for variations of shear strength parame-ters versus gasoil contamination obtained from triaxialtests are given in Figures 16 and 17. It can be seen thatcohesion increases before 8% gasoil and then, decreases;

Figure 16. Variations of cohesion versus gasoil content intriaxial tests.


Figure 17. Variations of friction angle versus gasoilcontent in triaxial tests.

but, friction angle has an inverse trend. Therefore,increase in gasoil up to 8% leads to increase in cohesion;but, when gasoil increases to 16%, cohesion decreases.This e�ect of gasoil on cohesion may relate to chargedstructure of clay plates with gasoil. Adding �ne sand topure clay increases the friction angle, because the sam-ple texture changes from quite plastic mode to a mix-ture with grained materials. In Figure 13 contour plotof friction angle indicates that the friction angle of pureclay in the condition of optimal compaction is 25.4 de-grees. Adding 10% and 20% �ne sand changes thisangle to 27.3 and 27.85 degrees, respectively. Gasoil ingeneral has decreasing role in friction angle. This role isbecause of the lubricating e�ect of gasoil that covers theexternal surfaces of the particles like a thin oil �lm andreduces the friction and conict between the particles.Increasing gasoil content up to 8% decreases the frictionangle to 11.72 degrees, while increasing gasoil contentup to 16% raises the friction angle to 21.93 degrees.Based on Figures 16 and 17, the important occurrenceis that the point of change in trend of variations offriction angle is exactly the same as that for cohesionvariations. The interpretation of this needs more de-tailed studies on interaction between clay, sand, water,and gasoil particles as well as the physicochemicalbehaviors of microstructures in tested specimens.

5.4. Direct shear testsChanging trends of friction angle and cohesion as wellas impact mechanisms of gasoil on cohesion and frictionangle have compatibility and strong resemblance withthe results of triaxial tests, but the amounts andquantities of the parameters are di�erent. This factrelates to the mechanism and structure of the shear boxthat impose a certain failure surface and rigid boundaryconditions on samples in comparison with the triaxialapparatus.

6. Summary and conclusions

To investigate the engineering properties of �ne soils

contaminated by gasoil, a laboratory testing programbased on the RSM was carried out. A type of pure clay(kaolinite), namely CL, in the uni�ed soil classi�cationsystem and clay mixed with di�erent amounts of �nesand (SP in USCS) were used as the �ne materialsin this study. The gasoil was added to specimenswith di�erent weight percentages up to 16% accordingto software selection. The samples were entirelymixed with gasoil and cured for 7-10 days in roomtemperature inside sealed plastic bags.

Plasticity indices as well as the density andcompaction characteristics of the specimens were mea-sured and shear strength parameters of the samplesprepared at 90% relative density were evaluated andinvestigated. Four models were developed for the tests.Based on the results, di�erent conclusions were driven,the most important of which are as follows:

Both the Plastic Limit (PL) and Liquid Limit(LL) of �ne soils are reduced as the gasoil contentincreases up to 16%. The rate of decreasing for pureclay soils is slightly higher than that for sandy claysoils;

In constant compaction e�ort, as the amount ofthe gasoil contamination increases (up to 16%), themaximum dry density increases and the optimummoisture content decreases. These variations forpure clay soils are more considerable than those forsandy clay soils;

The shear strength parameters of �ne soils exhibitopposite variations with increase in the gasoil con-tamination. While cohesion of the samples increasesup to 8% gasoil content, after which it startsdecreasing, internal friction angle decreases up to 8%gasoil content, after which it start increasing. Theseadverse e�ects of gasoil content may be attributedto the di�erence of shear parameters in nature; inone parameter, chemical reactions play the decisiverole while in the other, the mechanical behavior ofthe particles is determinative;

In this study, by using RSM and designing somegeotechnical tests, the engineering characteristicsof contaminated �ne soils were obtained with afewer number of tests. Also, di�erent equationswere developed for predicting the main geotechnicalparameters of clayey soils contaminated with gasoilin terms of �ne sand and gasoil contents of the layers.

References

1. Fine, P., Graber, E.R., and Yaron, B. \Soil interactionswith petroleum hydrocarbons abiotic processes", SoilTechnology, 10(1), pp. 133-153 (1997).

2. Pathak, P., Singh, D.N., Asce, F., Pandit, G.G., andApte, P.R. \Establishing sensitivity of distribution


coe�cient on various attributes of a soil contaminatedsystem", Journal of Hazardous Toxic and RadioactiveWaste, ASCE, 18, pp. 64-75 (2014).

3. Hong, S., Zhengtao, H., Xingang, L., and Guozhong,W. \Inuence of soil and hydrocarbon properties onthe solvent extraction of high-concentration weatheredpetroleum from contaminated soils", EnvironmentalScience and Pollution Research, 21(9), pp. 74-84(2014).

4. Puri, V.K. \Geotechnical aspects of oil-contaminatedsands", Soil and Sediment Contamination, Taylor &Francis, 9(4), pp. 359-374 (2000).

5. Rahman, Z.A., Hamzah, U., Taha, M.R., Ithnain,N.S., and Ahmad, N. \Inuence of oil contaminationon geotechnical properties of basaltic residual soil",American Journal of Applied Sciences, 7, pp. 954-961(2010).

6. Gupta, M.K. and Srivastava, R.K. \Evaluation of engi-neering properties of oil- contaminated soils", Journalof the Institution of Engineers (India), 90, pp. 37-42(2010).

7. Jia, Y.G., Wu, Q., Shang, H., Yang, Z.N., andShan, H.X. \The inuence of oil contamination onthe geotechnical properties of coastal sediments in theYellow River delta", Bulletin of Engineering Geologyand the Environment, 70(3), pp. 517-525 (2011).

8. Abousnina, R.M., Manalo, A., Shiau, J., and Lokuge,W. \E�ect of light crude oil contamination on thephysical and mechanical properties of �ne sand", Soil& Sediment Contamination: An International Journal,8(24), pp. 833-845 (2015).

9. Ojuri, O. and Epe, G. \Strength and leaching char-acteristics of crude oil contaminated sandy soils sta-bilized with sawdust ash-cement", Geo-Chicago 2016:Sustainable Waste Management and Remediation,Chicago-Illinoise, pp. 582-590 (2016).

10. Cook, E.E., Puri, V.K., and Shin, E.C. \Geotechnicalcharacteristics of crude oil-contaminated sands", Proc.of the 2nd Int. O�shore Polar Eng. Conf., San Fran-cisco, USA, pp. 384-387 (1992).

11. Khosravi, E., Ghasemzadeh, H., Sabour, M.R., andYazdani, H. \Geotechnical properties of gas oil-contaminated kaolinite", Engineering Geology, Else-vier, 166, pp. 11-16 (2013).

12. Pinedo, J., Ibanez, R., Lijzen, J.P.A., and Irabien, A.\Assessment of soil pollution based on total petroleumhydrocarbons and individual oil substances", Journalof Environmental Management, 130, pp. 72-79 (2013).

13. Khan, S.A., Tan, S.J., Abdul Rahman, E.K., andMuneerah, Dk.N. \E�ect of hydrocarbon contami-nation and subsequent remedial treatment on theengineering properties of soil", Proceeding of the 5thBrunei International Conference on Engineering andTechnology (BICET 2014), Brunei (2015).

14. Al-Sanad, H.A., Eid, W.K., and Ismael, N.F.\Geotechnical properties of oil contaminated Kuwaitisand", Journal of Geotechnical Engineering, ASCE,121, pp. 407-412 (1995).

15. Al-Sanad, H.A. and Ismael, N.F. \Aging e�ect on oilcontaminated Kuwaiti sand", Journal of Geotechnical& Geoenvironmental Engineering, ASCE, 123, pp.290-293 (1997).

16. Khamehchiyan, M., Charkhabi, A.H., and Tajik, M.\E�ects of crude oil contamination on geotechnicalproperties of clayey and sandy soils", EngineeringGeology, Elsevier, 89, pp. 220-229 (2007).

17. Nazir, A.K. \E�ect of motor oil contamination ongeotechnical properties of over consolidated clay",Alexandria Engineering Journal, 50, pp. 331-335(2011).

18. Kermani, M. and Ebadi, T. \The e�ect of oil con-tamination on the Geotechnical properties of �ne-grained soils", Soil and Sediment Contamination: AnInternational Journal, Taylor & Francis Group, 21, pp.655-671 (2012).

19. Liu, Z.B., Liu, S.Y., and Cai, Y. \Engineering propertytest of kaolin clay contaminated by diesel oil", Journalof Central South University China, 22(12), pp. 4837-4843 (2015).

20. Nasehi, S.A., Uromeihy, A., Nikudel, M.R., and Mor-sali, A. \Inuence of gasoil contamination on geotech-nical properties of �ne and coarse-grained soils", Jour-nal of Geotechnical and Geological Engineering, 34(1),pp. 333-345 (2016).

21. Gomez, F. and Sartaj, M. \Optimization of �eld scalebio-piles for bioremediation of petroleum hydrocarboncontaminated soil at low temperature conditions by re-sponse surface methodology (RSM)", Int. Biodeterior.Biodegrad, 89, pp. 103-109 (2014).

22. Kalali, A., Ebadi, T., Rabbani, A., and Sadri Moghad-dam, S. \Response surface methodology approach tothe optimization of oil hydrocarbon polluted soil reme-diation using enhanced soil washing", Int. J. Environ.Sci. Tech., Springer, 8(2), pp. 389-400 (2011).

23. Montgomery, D.C., Design and Analysis of Experi-ments, 4th. Ed., John Wiley and Sons, USA (1996).

24. Myers, R.H. and Montgomery, D.C., Response SurfaceMethodology: Process and Product Optimization UsingDesigned Experiments, 2nd. Ed., John Wiley and Sons,USA (2002).

25. Singh, S.K., Srivastava, R.K., and John, S. \Settle-ment characteristics of clayey soils contaminated withpetroleum hydrocarbons", Soil Sediment Contamina-tion, 17, pp. 290-300 (2008).

26. Mehrabi, N., Soleimani, M., Shari�fard, H., andMadadi Yeganeh, M. \Optimization of phosphate re-moval from drinking water with activated carbon usingresponse surface methodology (RSM)", Desalinationand Water Treatment, Taylor & Francis, 57(33), pp.15613-15618 (2015).


27. Albert, T.Y. \Di�uze double-layer equations in SIunits", J. of Geotech. Eng., ASCE., 118(12), pp. 2000-2005 (1992).

28. Di Matteo, L., Bigotti, F., and Ricco, R. \Com-pressibility of kaolinitie clay contaminated by ethanol-gasoline blends", J. of Geotech. & Geoenvr. Eng.,ASCE, 137, pp. 846-849 (2011).

29. Sridharan, A., El-Shafei, A., and Miura, N. \A studyon the dominating mechanisms and parameters inu-encing the physical properties of Ariake clay", Inter-national Association of Lowland Technology Journal,2, pp. 55-70 (2000).

30. Olgun, M. and Y�ld�z, M. \E�ect of organic uids onthe geotechnical behavior of a highly plastic clayeysoil", Applied Clay Science, Elsevier, 48, pp. 615-621(2010).

31. Kaya, A. and Fang, H.Y. \Experimental evidence ofreduction in attractive and repulsive forces betweenclay particles permeated with organic uids", Cana-dian Geotechnical Journal, Technical Note, 42, pp.632-640 (2005).

32. Meegoda, N.J. and Ratnaweera, P. \Compressibility ofcontaminated �ne grained soils", Geotechnical TestingJournal, 17(1), pp. 101-112 (1994).

33. Ur-Rehman, H., Abduljauwad, S.N., and Akram,T. \Geotechnical behavior of oil-contaminated �ne-grained soils", Electronic. J. of Geotech. Eng., 12A,pp. 15-23 (2007).

Biographies

Fatemeh Mir Mohammad Hosseini Completedher BSc in Civil Engineering Faculty of AmirkabirUniversity of Technology, and her MSc in Environmen-tal Faculty of Tehran University. At present, she isthe PhD student of Environmental Group in Civil &Environmental faculty of Amirkabir University. Herresearch topic in PhD studies is investigating the static

and dynamic properties of �ne cohesive soils contami-nated by gasoil by means of extensive monotonic andcyclic testings.

Taghi Ebadi is an Assistant Professor in Environ-mental Group of Civil and Environmental faculty ofAmirkabir University of Technology. He received hisPhD (1376) from Concordia University in Canada andhis MSc (1370) from Amirkabir University of Technol-ogy. His main research interests are geo-environmentalstudies by emphasizes on the laboratory and �eldinvestigations of contaminants on soils.

Abolfazl Eslami is a Professor in Geotechnical groupof Civil and Environmental Faculty of Amirkabir Uni-versity of Technology. He received his PhD (1375)from Ottawa University in Canada and his MSc (1367)from Amirkabir University of Technology. His mainresearch interests are soil bearing capacity studies byemphasizes on the soil improvements methods andtechniques.

Seyed Majdeddin Mir Mohammad Hosseini is aProfessor of Geotechnical Group of Civil and Environ-mental Faculty of Amirkabir University of Technology.He received his PhD (1366) from Leeds Universityin UK and his MSc (1355) from Tehran University.His main research interests are soil dynamics andgeotechnical earthquake engineering with focusing onlaboratory works.

Hamid Reza Jahangard Completed his BSc in CivilEngineering Faculty of Tehran University. He did hisMSc studies in Civil and Environmental Faculty ofAmirkabir University of Technology and carried out hisMSc research project on investigation of cyclic strengthof �ne soils contaminated by gasoil.

Investigation into geotechnical properties of clayey soils...

Documents

Transcript of Investigation into geotechnical properties of clayey soils...